Modeling HSGPS Doppler Errors in Indoor Environments for Pedestrian Dead-Reckoning

The use of high sensitivity GPS (HSGPS) receivers integrated with dead-reckoning sensors for pedestrian navigation has been broadly investigated and applied in the past decade. Pedestrian dead-reckoning (PDR) generally involves heading determination, step detection, and step length updates (Lachapelle et al 2006). Whenever HSGPS measurements are available, PDR parameters can be calibrated by making use of these measurements. However, the accuracy of the integrated system is governed by the quality of HSGPS measurements. Indoor HSGPS performance in wooden or concrete buildings has been assessed by numerous researchers (e. g. Kjaergaard et al 2010). Variance of Doppler estimates under standard and block processing techniques were analyzed and verified in Borio et al (2010). This paper provides theoretical results to characterize the Doppler errors obtained using maximum likelihood estimate (MLE) algorithms in various indoor environments with moderate pedestrian motion dynamics (for example, velocities are assumed below 2.5 m/s). Combined LOS/NLOS signal models are considered. An essential parameter set including Ricean factor, averaged multipath azimuth angle, multipath azimuth beamwidth, averaged multipath elevation angle, multipath elevation beamwidth, LOS azimuth, and LOS elevation angle is defined, in order to quantify the Doppler errors of MLE algorithms. Two real data sets were collected in indoor environments. MLE Doppler biases are observed in certain directional multipath dominant environments, which are in agreement with theoretical simulation results.